Re: On schema quality and schema limitations

Le mar 13/04/2004 à 15:31, Dominique Hazaël-Massieux a écrit :
> But in the given case, the English prose specification pre-exists to the
> formal one ; and the formal specification is in fact mainly used for
> validation and formalization purposes ; more generally, given that the
> expressive power of English is much larger than the expressive power of
> any formal language, you cannot necessarily transform all the
> requirements you want to set in formal language requirements ; in
> particular, there may always be cases that cannot be expressed in the
> formal language that would be refined by the prose.

Somewhat related to this discussion, I've applied one of the possible
usages of formal languages in specification to my review of the
VoiceXML2.1 [1], where I was able to find a buggy example in the spec by
extracting all the XML examples (using an XSLT [2]) and validating the
extracted files; note that in this case, I could only check well-formed
since the DTD and XML Schema aren't available yet, but I think this
approach could be applied to other specifications with interest. See my
blog entry on the topic [3].

Dom

1. http://lists.w3.org/Archives/Public/www-voice/2004JulSep/0009.html
2. http://www.w3.org/2004/07/extract-examples.xsl
3.
http://people.w3.org/~dom/archives/2004/07/extracting-examples-from-a-specification/
-- 
Dominique Hazaël-Massieux - http://www.w3.org/People/Dom/
W3C/ERCIM
mailto:dom@w3.org