Re: RDF Core test driven development and QA Test Doc

I think the comments interspersed below are broadly supportive of the 
point you are making.

Jeremy Carroll wrote:
[...]

> The comment currently reads:
> 
> [[
> Functional Analysis and Test Driven Development
> 
>  At times, RDF Core used a test driven specification development methodology. 
> 
> Issues from the issue list were resolved by agreeing test cases.

Thats right, though I'd modify the emphasis.  Test cases were used to 
express/represent/communicate issues and decisions.  Test cases were 
used to explain/communicate clearly what the issue was, and what the 
resolution was.  In some cases this worked wonderfully well.  For 
example, the semantics example you quote had caused no end of confusion 
before it was expressed as a test case.

It was really the issue we were deciding, not the test case per se.  The 
representation of the issue and the decision was one or more test cases. 
  The test cases were a means to an end, not an end in themselves.  We 
were not setting out to produce a conformance test suite, though as a 
side effect of the process, we got a useful, though incomplete, set of 
test cases for software.

> 
> The editors then had complete freedom to write text which conformed with the 
> test cases. (The text was later reviewed, so the freedom was not as excessive 
> as it seems). 

Well, as you point out, 'complete freedom' is overstating it.  The WG 
relied on the skill of the editors to draft words for the specs that 
expressed the WG's decision.  Generally, this draft text was reviewed by 
the WG.  Usually the test cases were retained in the test suite to 
augment the specifications.

I think this worked well for us.

[...]

 >
 > In as much as the Test Guidelines and the QAF prohibit and/or 
obstruct this
 > behaviour

Raised eyebrows!  This document is restrictive?  It is saying that a WG 
is not allowed to do something, rather than advising on good practice?

[...]

 > I note that at your telecon on Monday, Sandro will be presenting 
stuff about
 > WebOnt and RDF Core approach to testing. I am available to attend if you
 > would like. .
 >
 > Sandro is likely to be friendlier than me, and I suspect has seen 
enough of
 > how both groups operate to adequately inform you.

With no disrespect to Sandro, (we are deeply indebted to him for 
automating the analysis of test results) he was not a member of RDFCore. 
  If input on how RDFCore actually used test cases in spec development 
is sought, if may well be useful also to have someone present who 
participated more broadly in the process.

Brian

Received on Friday, 2 January 2004 18:07:59 UTC