[XML-SIG] "Borrowed" tests
Uche Ogbuji
uche.ogbuji@fourthought.com
Fri, 06 Apr 2001 09:45:22 -0600
> On Thu, 5 Apr 2001, Uche Ogbuji wrote:
> > I've long had the practice of placing test cases based on bug reports or
> > examples provided by others into a "test_suite/borrowed" directory. I think
> > this is useful, and important to distinguish from the "canned" tests.
>
> Your logic here escapes me. Why would this distinction
> be necessary?
It's not necessary. My main goal is to get the tests back in regardless of
where they go. I was just suggesting a course based on what I'd done before.
The reason why I made the distinction before is that the "canned" tests were
grouped according to speficication fiat, so there'd be, say a test_variable.py
which tests the various diktats of section 11.4 of the XSLT spec.
The tests that come from bug-reports and borrowed code tend not to be so easy
to neatly categorize. Therefore I placed them in a separate directory just to
provide a separate axis of grouping. Not a big deal, except that this
directory got lost for 4DOM as it moved to PyXML.
> It just seems like extra work with no
> additional benifit.
There's no extra work whatsoever. Why do you think there is?
> As I find bugs in my software I
> add it to my regression test, and usually name the
> test after the bug. You could opt for a naming
> convention, where "canned" tests start with a "c"
> and borrowed tests start with a "b" (for bug)
This is exactly what I do. The *only* difference being that I differentiate
by directory placement rather than prefix. I actually think the directory
approach is less work than the prefix approach, but perhaps we're
misunderstanding each other.
--
Uche Ogbuji Principal Consultant
uche.ogbuji@fourthought.com +1 303 583 9900 x 101
Fourthought, Inc. http://Fourthought.com
4735 East Walnut St, Ste. C, Boulder, CO 80301-2537, USA
Software-engineering, knowledge-management, XML, CORBA, Linux, Python