Re: lack of testability definition

On Tue, 4 Nov 2003, Bjoern Hoehrmann wrote:

> * Alex Rousskov wrote:
> >On Tue, 4 Nov 2003, Bjoern Hoehrmann wrote:
> >
> >> http://www.w3.org/TR/wcag2-tech-req/ has a nice definition.
> >
> >Note that the definition you quote renders pretty much all behavioral
> >requirements such as "agent MUST do X" Machine Untestable. Since most
> >of those requirements are also not possible to test via humans, the
> >approach renders pretty much all behavioral specs as Machine and Human
> >Untestable.
>
> Sorry? Could you give some example requirements?

Sure. Pretty much every MUST from HTTP (RFC 2616) or any other
behavioral spec. For example,

	Proxy MUST delete Connection headers.
	Proxy MUST NOT forward hop-by-hop headers.
	Cache MUST NOT cache responses with Cache-Control: private.
	Cache MUST NOT store responses with Cache-Control: no-store.
 	An agent MUST NOT send chunked encoding to HTTP/1.0 agent.

and so on...

None of the above are Machine Testable given
http://www.w3.org/TR/wcag2-tech-req/ definition:

	Definition: Machine Testable: There is a known algorithm that
	will determine, with complete reliability, whether the
	technique has been implemented or not. Probabilistic
	algorithms are not sufficient.]

HTH,

Alex.

Received on Tuesday, 4 November 2003 13:16:50 UTC