[trustable-software] We've got this all wrong !

trustable at panic.fluff.org trustable at panic.fluff.org
Mon Apr 16 16:17:21 BST 2018


On Mon, 16 Apr 2018, Paul Sherwood wrote:

> Hi Edmund
> On 2018-04-16 10:12, trustable at panic.fluff.org wrote:
>> So I'm arguing we NEVER have requirements, we do have intents, we
>> accept what the tests tell us about behaviour and use this as a RISK
>> model to see if we are heading somewhere.
>
> You're not using the t.notation so I'm not sure what you now mean by 
> 'requirements'.

      I'm carefully avoiding the t.notation in this debate.
      One problem is how we definite what a requirement is
      within the t.notation form we never put formal
      definitions on their value. Only on their relationships


>I'm kind of hoping you are right, but what about:
>
> 1) Re-development of existing solution using existing requirements from, say, 
> DOORS - which do have tests, albeit retrofitted

  I'm saying I think though there is a lot of employment in collecting 
these requirements, in fact we only know the behaviour and reliance based 
on the test harness and results. These are then retrofitted to the 
documentation. What's in fact worse about this is we don't retrofit a risk 
measure to the results. We simply assume because the tests mirrored 
expected behaviour and were reproducible we have 'trustable' software.


> 2) Project which must satisfy a specific law (e.g. reversing camera video 
> must be available by X seconds)

So again I'm saying we begin with an 'Intent' we build tests to confirm 
this with the auditor for the  specification. We don't in fact start with 
this in mind.. The regular statement being JFDI and we'll iteratively 
improve later.


I think this might even apply further and in fact there is no 
architecture, only a bunch of purchasing decisions and so career 
progression decisions. In fact the architecture is a result of the 
implementation and the tests and only exists after the implementation.

The modern IaaS & PaaS approaches have value because of the easy to change 
your decisions very late in the day, and the cost isn't the physical. How 
often do we do product comparisons for solutions which turn out simply to 
be a scoring system to choose what we already know or need on our CVs 
without in fact looking for a more complete solution.


Edmund
-- 
========================================================================
Edmund J. Sutcliffe                     Thoughtful Solutions; Creatively
<edmunds at panic.fluff.org>               Implemented and Communicated
<http://panic.fluff.org>                +44 (0) 7976 938841




More information about the trustable-software mailing list