Some discussion additions to Wiki

[stephen.green@systml.co.uk]

I just added the following to the "What a TA should and should not be"
not sure if that is the best place for it though
(they also relate to a comment we received from Dave Martson on the
comment list regarding automation of producing tests from TAs
http://lists.oasis-open.org/archives/tag-comment/200708/msg00001.html)

"# Is there a need to distinguish the following in the TA model?:

     * Prose TA copied exactly from a specification
     * Prose TA not copied exactly from a specification
     * A predicate - a more formal expression than free-text prose and  
which evaluates to true or false
     * A structure to allow separation of prose and/or predicate  
expressions which describe preconditions for tests, post conditions  
for tests, triggers and outcomes (stephen)

# If there are parts of the model which allow, mandate or recommend  
predicate expressions would we expect, cater for and recommend the use  
of a constraint language for these predicates

     * if they are expressed as XML would such XML be either
           o referred to as links, XQueries or IDs in the TA document
           o included in the TA document
     * what would be suitable for such expressions?
           o UML's Object Constraint Language (OCL) or OCL XMI (XML  
representation of OCL)?
           o Web Ontology Language (OWL)?
           o Semantic Web Rule Language (SWRL)?
           o Rule Markup Language(s)?
     * are XML-based expressions to be preferred to non-XML (such as  
Backus-Naur based expressions)? (stephen)

# Suggestion:

     * we have somewhat divergent requirements for simple, mainly  
prose TAs versus more complex, structured and perhaps fully  
automatable TAs
     * does this lead perhaps to two possible profiles - a profile  
aimed at human-readable documentation of TAs and another profile aimed  
at being either formally processed according to a particular notation  
or methodology or even a set of TAs which are so defined (as  
predicates or sets of predicates using say OCL XMI) as to be  
processable into tests automatically with software tools (stephen)

# If the predicates of the structured part of the TA model can be  
expressed with a language such as OCL then that language (as OCL does)  
may have its own represenations of what it calls preconditions and  
postconditions

     * does this mean there might be duplication or confusion with  
what we have modeled using the same terms?
     * are these actually the same things or do we distinguish ours as  
being in a different layer with different purposes or meanings (
           o are ours preconditions and postconditions constraining  
something which OCL's say aren't (such as constraints more or whether  
the test should occur rather than on what the test itself) or is there  
such a fine distinction as to just be a grey area which might at worst  
confuse or at least add the possibility of redundancy? (stephen)"


-- 
Stephen Green

Partner
SystML, http://www.systml.co.uk
Tel: +44 (0) 117 9541606

http://www.biblegateway.com/passage/?search=matthew+22:37 .. and voice