RE: [tag-discuss] the role of an XML mark-up for TAs

[david_marston@us.ibm.com]

In my view, each test case may specify four phases:
1) setup
2) execution
3) comparison of actual result vs. expected, producing a test "outcome"
4) cleanup

Carmelo and I are contending that (1) should be considered broadly
enough so that each test case can be run independently, and you can
apply filtering on the set of test cases in any way you need, without
worrying about sequencing of tests. Any one test can even be run alone.
That doesn't say anything about TAs, really. We are also assuming that
(2) produces something concrete enough for (3) to operate on.

The implication is that TAs become the basis for both phases 1 and 2.
Phase 1 includes setting up preconditions for a test case other than
those which should or must be effected in Phase 2. Phase 1 could also
include generic setup.

Phase 4 would typically include generic cleanup, but could include
post-conditions specific to a test case. We should debate what subset
of those are in scope for a TA design.

AB>Can we have a TA for testing Delta?

This doesn't strike me as being in scope. I think a TA can only say
that if a particular set of conditions arises, some action produces
some result. How the conditions arise should be independent of any
"delta" situation as far as I can see, or the "delta" may be one
way to cause a set of conditions. Can you elaborate?
.................David Marston