[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]
Subject: TA Model
I don't like restricting the flow part of the model to just conditions, trigger and result. I question whether there is always a trigger as such (although in some test technologies there might always be one, but we are to be independent of that I think). I question whether there is always *only* a trigger and result. In the UBL work there is absolutely nothing in the spec to specify a trigger or a result, at least not normatively. I therefore feel there is a need to specify a 'test' which can be standalone in the TA. example: An invoice has the following element as its document root 'Invoice'. Or even An invoice MUST have the following element as its document root 'Invoice'. Either seems appropriate since the latter could be exact wording from the spec (but I'm over-selling that argument I fear, sorry). Secondly, during the running of a test there may be a need to watch for certain things which are not exactly the result as such. So the extra 'test' element could assert the need to watch for these things and perhaps terminate the test if they occur or assert that they have an impact of some sort on the result or the verdict irrespsctive of the result. Typically this test element would be stated as a predicate but perhaps not necessarily (to allow for all kinds of test methodologies and test assertion authoring methodologies). Perhaps an attribute in each element (trigger, etc) could indicate whether the content evaluates to a boolean or there could be some such thing to express any constraints on the content (which would align with the UML Testing Profile which has the ability to add such constraints). -- 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
[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]