RE: [tag-discuss] Terminology

[david_marston@us.ibm.com]

I used the term "test specification" in a deliberately loose way. 
Sometimes a TA (or combination of TAs) will say everything you need to 
write a test case. In other situations, the TAs will "specify" part of 
what the test case needs to do, establishing boundaries but not dictating 
the whole case. Test specifications are used by a test case writer to 
write a case. "Test plans" are used by a test case writer to scope out a 
set of test cases, each different, that exercise some TAs or, lacking TAs, 
a set of testable sentences. A test plan contains test specifications for 
several test cases. While writing test plans and test cases, software QA 
people would also produce test case metadata to facilitate administration 
of the tests. (As mentioned previously, the metadata may include pointers 
to TAs.)

Test Assertions (TAs) are in the realm of documentation and editorial 
people, who are writing a specification that describes the behavior of 
some product (software, hardware, data, whatever). I think the TA TC, if 
convened, would be especially interested in designing a format for TAs 
that would work for "standards" organizations like OASIS, where the specs 
define the behavior of a class of products, implemented by several vendors 
and substitutable for one another. These specs, with or without TAs, are 
used by implementers writing code. Implementers doing Test-Driven 
Development would be especially appreciative of having TAs before they 
begin coding.

Thus, TAs are produced by the TC or WG producing the prose spec, but they 
are not for the sole use of software QA people. A TC or WG may issue TAs 
without issuing any test cases.
.................David Marston