- From: (unknown charset) Alex Rousskov <rousskov@measurement-factory.com>
- Date: Fri, 9 Apr 2004 09:37:52 -0600 (MDT)
- To: (unknown charset) Dominique Hazaël-Massieux <dom@w3.org>
- Cc: (unknown charset) www-qa@w3.org
On Fri, 9 Apr 2004, Dominique [ISO-8859-1] Hazaël-Massieux wrote: > worth trying to solve in W3C as a quality process: how to check that the > formal language (defined in schema, dtd, ...) matches the > English-written specification ; matching has several aspects: > * does the formal language add constraint not expressed in the spec? > * does the formal language fail to check constraints expressed in the > specification? > * does the formal language contradicts the specification? > > I think the second question should be recast into "does the formal > specification uses all its expressive power to match the constraints > set in the specification?", which is one of the issues that you > raise in your comments. I believe going down the path outlined by the above questions may be a mistake. Since formal language is formal and English language is not, it is not possible to establish their formal equivalence without converting English to something formal (which would defeat the original purpose). A good specification must not duplicate formal-language requirements in English prose. For example, a document may say "An agent MUST reject any message that violates the message format specified by the above BNF", but duplicating BNF requirements using English is not a good idea. If a specification is written without the use of a formal language and some other document rewrites the specification using formal language, the formal equivalence of the two documents cannot be estabslished. In other words, a good specification will not cause the questions you are trying to answer. Those questions probably do not have an answer. Alex.
Received on Friday, 9 April 2004 11:38:08 UTC