Re: [SpecGL Draft] E. Good Practice: Write Test Assertions

On Friday 16 July 2004 10:50, Karl Dubost wrote:
> 
> Le 16 juil. 2004, à 04:43, Andrew Thackrah a écrit :
> >> Do not put the feature in the specification without having its
> >> corresponding test assertion(s).
> >
> > Are you suggesting that assertions are written before the 
> > specification?
> > If so I don't this ever happens (?) I thought assertions were derived 
> > from
> > specs.
> 
> At the same time, you can certainly write few test assertions at the 
> time you are writing the specification. For example for SpecGL Lite, 
> the template shows that a Principle or a Good Practice *could* be 
> considered as a test assertion or at least help to create one, as 
> exactly you would do with test cases. If you propose a feature to add 
> to a technology and you can't write the test assertion which goes with 
> it, it means that there is already a problem in the way you have 
> conceived, designed or written the feature.
> 
> Though it doesn't mean that WG is done when the specification has been 
> finished.
> 
> Why do you think it's not possible to write a test assertion at the 
> time you are writing the specification?

Hi Karl - I didn't say that! Or rather, 'I never made that assertion' :-)

It is of course possible, but the suggested text does not communicate the idea 
as clearly as you have communicated it in your reply to me.

Currently the text suggests that an assertion is prerequisite to a feature 
description (well, that's how I read, and may be others?).  How about this 
instead:

"Try to write assertions as you add features to a specification.  If you can't 
write the test assertion for the feature it suggests that there is a problem 
in the way you have designed or explained the feature."

cheers,
 -Andrew

Received on Friday, 16 July 2004 06:01:24 UTC