[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]
Subject: Notes of Best Practice comments
Notes taken about Best Practice sections from today's half of TC meeting: (elaborations or amendments welcome) BEST PRACTICE GUIDANCE FOR WRITING TEST ASSERTIONS -GENERAL PRINCIPLES --THE VALUE OF WRITING TEST ASSERTIONS Add that the writing of TAs provides for a tighter spec Add that the TA is ideally written in parallel with spec rather than merely early on in writing the spec This is all part of the foundation rather than best practice --WHO WRITES TEST ASSERTIONS Clarify mention of TA mandating in relation to mandating of conformance clauses (Relates to the way a conformance clause can refer to a set of TAs) The mandating aspect refers to mandating of the inclusion of TA creation in the process of spec writing --THE RELATIONSHIP BETWEEN TEST ASSERTIONS AND CONFORMANCE Conformance clauses are at a higher level. The conformance clauses will likely include lots more than the TAs, such as specifying which parts of the specification are normative or informative, the use of keywords such as MUST, SHOULD, MAY, etc. In contrast the TAs are a refinement to of parts of specification to provide the foundation for building tests TAs singular / Conformance clauses groups TAs have a narrower purpose Conformance clauses may make ref to TAs but are distinct from them Conformance clauses may specify conformance to levels, say, in relation to different TAs TAs like leafnodes in branching structure For examaple, Sun's conformance clauses (called compatibility requirements) run into pages and include details about which parts (or all) of the specifcation to implement, whether subsets or supersets may be implemented whether the Sun test harness must be used. Patrick not sure about just adding this section as a definition to the Glossary - more of a fundamental matter to be defined up front --TEST ENVIRONMENT CONSIDERATIONS TA should NOT say how to test (the TA writers may not know how to test at time of writing) but should give enough detail so can tests can be constructed Again, include this in fundamentals section up front Debatable whether the IUT should be required at TA level - either can be part of conformance or may be generalized for a set of TAs For example, the conformance clauses might list classes of implementations or IUTs and this might result in production of an enumerated list or controlled vocabulary of IUTs (either in the conformance clause or test assertion suite introduction or both) which can be referred to, perhaps by coded values, in the individual TAs A best practice technique is to list the IUTs to which a TA applies using words like 'applies to ...' Consensus that IUT should not be required as part of the individual TA There may be granularity issues Use of 'how': not to be used regarding control of test but more appropriate in the behaviour aspect of a TA Harness: metadata bridges the gap between the TA and a test case in deterministic way --COVERAGE Patrick will cover this -WRITING SIMPLE STRUCTURE TEST ASSERTIONS Care needs to be excercised when converting a requirement to a TA not to be adding interpretation or subtly changing the meaning Just using references to spec avoid adding interpretations but then care needs to be taken to make the reference sufficiently specific such as by specifying the boundary of which part is referenced or to reference a sentence There is value in including text outside the spec in the TA does but there is also a danger of divorcing the text from its context, formatting, etc One best practice is to use some technique to highlight text actually in the spec itself and add a TA Id to it there Note that a the relationship between TA and spec may be many-to-many; there may even be more than one spec to which references need to be made -MORE COMPLEX STRUCTURE TEST ASSERTIONS [look at later] -AUTOMATION Note: generation from schema can just go straight to test by-passing TA -IMPACT OF VARIABILITY IN SPECIFICATIONS ON TEST ASSERTIONS e.g. 'applies to' -GROUPING -SEQUENCING Patrick still nervous - prob if modular specs on top of other specs - 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]