Re: [Oorexx-devel] New ooRexx Test Discussion

[Mark M. <mie...@gm...>]

On Mon, Sep 8, 2008 at 1:15 PM, Rick McGuire <obj...@gm...> wrote:

> There will be a lot of variations of the above driven mostly by the
> ooRexx test code.
>
> I think this strategy of having a simple native stub and the actual
> test code being primarily ooRexx code will work well for a lot of
> these APIs.  The stub will be a simple API call and the real data will
> be provided by the ooRexx variations.

This is also what I pictured, although I hadn't got around to the
actual details.

Having a simple native stub that can be exercised in numerous ways has
these advantages:  The bulk of the actual tests are implemented within
ooTest, taking advantage of the infrastructure we already have
developed.  Adding different edge cases as we think of them is
simpler.  Once the C++ code for the tests is written, it shouldn't
need to change much.  Which would mean that  precompiled binaries
become easier to do.

--
Mark Miesfeld