Re: Should Test Assertions be required

• From: <david_marston@us.ibm.com>
• Date: Fri, 6 Jun 2003 12:52:42 -0400
• To: www-qa@w3.org
• Message-ID: <OF44C60260.4870A841-ON85256D3D.005C563D@lotus.com>

Alex Rousskov writes:
>> 3. facilitates test generation
>True. However, it may facilitate generation of both inferior
>implementations and inferior tests because:
>...
>  - Implementors with little resources may focus on satisfying
>    (coding for) documented TAs instead of satisfying the
>    specification itself. A very common pitfall, IMO.

Good point. SpecGL should make it clear that the disclaimer section
disclaims the completeness of the TAs, just as it disclaims the
completeness of any test suite. Implementations must conform to the
entire applicable material of the spec in order to be considered
conformant.
.................David Marston

Received on Friday, 6 June 2003 12:52:52 UTC