On the basis of the discussion above I just committed revision 532749 with the following comment === I inserted an intermediate junit 3.8 style specific testcase class above DataObjectListTest (and propose to do this for others). The intermediate class delegates much function off to a non junit 3.8 specific class. So now any 4.1 style test classes can inherit from the non-junit superclass, and the junit 3.8 style ones can use the junit specific ones. This allows us to 1) promote setUp/tearDown to public to promote test harness agnosticism, 2) ensure that the testHelper has been initialized once per setUp() call, thereby permitting runs of individual tests, 3) do standard setup stuff such as create a HelperContext per test execution.
I also moved the DataObjectListTest to the Approved set === So this fixes some niggles centrally (cross talk due to HelperContext reuse, and running small sets/individual tests) and allows us to promote harness agnosticism while carry on with a hybrid approach to the junit style. We can readily switch classes between approaches if we decide to converge on a single approach at some time. I propose to adjust all test classes this way as i go through them, and eventually do away with the static TestHelper instance in the CTSSuite class. Kelvin. On 20/04/07, Frank Budinsky <[EMAIL PROTECTED]> wrote:
I agree. Frank. "Andy Grove" <[EMAIL PROTECTED]> wrote on 04/20/2007 01:14:51 PM: > > I would certainly prefer to continue with junit. > > There are frameworks such as cactus, that allow junit tests to be run in > J2EE environments, and if vendors need the ability to run the tests in > some other environment that is not supported by junit or cactus then > they always have the option of developing their own test runners or > tweaking the junit code to fit their requirements. This does seem like > an edge case and it would seem appropriate for those users to invest the > effort to solve the problem rather than putting an extra burden on > developing the general purpose CTS. > > Thanks, > > Andy. > > -----Original Message----- > From: kelvin goodson [mailto:[EMAIL PROTECTED] > Sent: 20 April 2007 17:19 > To: [email protected] > Subject: Re: [Java SDO CTS] Junit 4.1 pattern for calling setUp when > classes don't inherit from TestCase > > The Junit tooling is so useful I'd be loath to drop it as the harness > that the Tuscany implementation uses for exercising the tests. I'm going > to do a bit of playing to see what solutions are practical, but I'm > concerned that we may be considering putting significant effort into a > goal that's rather too theoretical, as junit seems so ubiquitous. > > Regards, Kelvin. > > On 20/04/07, Andy Grove < [EMAIL PROTECTED]> wrote: > <snip/> > > One option is to stop using junit completely and replicate the useful > > features in a minimal test framework that supports parameterized tests > > > e.g. we could introduce a CTSTestCase interface: > > > > > > <snip/> > > --------------------------------------------------------------------- > To unsubscribe, e-mail: [EMAIL PROTECTED] > For additional commands, e-mail: [EMAIL PROTECTED] > --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
