Re: On schema quality and schema limitations

Le sam 10/04/2004 à 01:34, Bjoern Hoehrmann a écrit :
> >Very interesting reading! I expect you'll turn most of its contents into
> >comments for the XHTML Modularization SE?
> 
> Hmm, maybe... I am not sure what to tell them.

Well, you could point them to the inconsistencies you've identified
between the schema and the specification (and also those that you detail
below).

> I would say that mail contains just more illustration for the issue,
> adding a link to the issue might be a good idea, I don't think there
> is anything new here.

OK, thanks for the clarification.

>   http://www.w3.org/TR/xhtml1/#h-4.11
> Now I do not know whether the target attribute falls into this category.
> It does not suffer from the case-sensivity problem in XML DTDs...

Hmm... The section you quote above reads "HTML 4 and XHTML both have
some attributes that have pre-defined and limited sets of values" ;
since target has pre-defined sets of of values, I guess it does fall in
this category.

> These do not co-exist very well... If multiple URIs are not allowed,
> Tidy should fix them... How would one review Schemas that are based on
> a contradictory specification that is not fixed?

I guess this should be part of your review to the Schemas :)

>  And how to figure out
> what to do for Tidy? A new data type? Special case the profile attribute
> to complain about spaces but not fixing them?

(my understanding of the spec is that it does allow whitespace separated
URIs, so spaces should be just OK ; given the very low usage of the
profile attribute, the chances that someone uses it with a non-escaped
space in it seems fairly low anyway).

> What I do have is 
>   http://www.websitedev.de/markup/validator/tests/
> which might be of some interest in this regard.

Most certainly, indeed. I'll try to see if I can come up with test cases
showing inconsistencies (no promise, though, I don't have much time on
my hands, and I'm far from being any expert in XML Schema...).

One idea to help testing at least this type of specification: generate
documents based on the different formal specifications (DTD, XML Schema,
Relax NG) and cross-validate them ; I don't know how hard this would be,
though...

Dom
-- 
Dominique Hazaël-Massieux - http://www.w3.org/People/Dom/
W3C/ERCIM
mailto:dom@w3.org