Re: Use case definition

Andrew et al --

I'd like to add something to the thread, in advance of us tackling the issue:

1.) the current wording of CK1.2 implies formalism for use cases, user 
scenarios, usage scenarios, or whatever we decide to call them.  But there 
are abundant examples where they are prose-y descriptions of sample user 
scenarios, examples, etc.

2.) and there is the question (issue #72) whether they are normative or 
informative.

I think that we (QAWG) believe that "SOAP Version 1.2" satisfies the two 
checkpoints, in particular section 2, 3, 4 of Part 0: Primer.  All of Part 
0 is informative, and it is mostly descriptive prose, with some actual 
SOAP-code examples accompanying the prose.

So maybe we need to split the checkpoints?  Allow informative, prose user 
scenario descriptions to satisfy the priority 1 checkpoints, and give 
"extra credit" (P2 or P3 checkpoints) for formalism and normativity 
(whatever the latter actually means)?

-Lofton.

At 04:27 PM 7/29/02 +0100, Andrew Thackrah wrote:

>Re Action Item A-2002-07-10-1 There is an outstanding issue (#72, [1]) on 
>the Spec. guidelines doc (07/22 version, [2]) relating to  checkpoint 1.2; 
>the definition of a use case and the implication of it being normative.
>
>  The problem is that there is no formal definition of what a use case is, 
> and that this undefined  thing is being made normative. Anything that is 
> normative in a spec. is important and so should  be defined carefully 
> before inclusion.
>   What is a use case? There are several definitions but I think that one 
> that is closest to the  SpecGL idea of a formal user scenario involves a 
> related sequence of actions or events,  quantified with standard inputs 
> and expected outputs.
>   In my experience a scenario is composed of a chain of lower level test 
> assertions. (The purpose of the  scenario is often to verify 
> interoperability, something not easily done at the 'atomic' level of an
>assertion.)
>  This implies the existence of low level, testable assertions being 
> available before a use case can  be defined - and this overlaps with 
> Guideline 15 and the idea of test cases.  So I think we need to answer 
> the question   "Is a Use Case the same thing as a Test case?"   If not, 
> how is it different?
>   To make a start at answering this I'll note that the UML style of use 
> case includes actors with specific  roles. The various 'users' to which 
> SpecGL scenarios might apply to may correspond to actors.  However, UML 
> use cases capture behaviour, not activity.
>   If a SpecGL use case captures behaviour, how is it formalised and tested?.
>  If a specGL use case does not capture behaviour - does it capture 
> activity instead? Or something else?
>
>  -Andrew
>
>  [1] http://www.w3.org/QA/WG/qawg-issues-html.html#x72
>  [2] http://www.w3.org/QA/WG/2002/07/qaframe-spec-0722.html
>

Received on Wednesday, 7 August 2002 20:06:59 UTC