Re: 'Re: "Re: [xml] schema patch"'

[Daniel Veillard <veillard redhat com>]

On Wed, Jun 09, 2004 at 11:02:27AM +0200, Kasimier Buchcik wrote:
> Hi,
> >   Hum, out of curiosity, how did you implement it ?
> > My take on this was that I would build a function to check if
> > an automata A was a subset of an automata B and then compare 
> > the compiled automata at the end of the Schemas compilation process.
> > I assume you used a different way
>
> The implemented sections 4.1 - 4.2 (restriction) and 1.3 (extension) 
> deal with attribute wildcards only - it's just a tiny piece of code. You 
> refer to the validity of derivation of the content model; I'm really 
> scared of this one ;-) Please take a look at [1].
[...]
> [1] http://www.kohsuke.org/xmlschema/XMLSchemaDOsAndDONTs.html#avoid_complex

  Sounds a lot like "Don't use W3C XML Schemas Structure" to me, which is
not something I would violently disagree with myself :-)
  Still automata subset checking is gonna be an interesting piece of code
if I ever get the time to do it...

  Also since you're the person pushing the most work in the Schemas code
those days, one of the future change will be support for streaming validation.
Just keep in the corner of your head that one day the code validation code
will have to be converted to handle a SAX2 event stream, and it's better
to not use the instance tree structure too much in the validation APIs
(that will be a major work anyway ...)

Daniel

-- 
Daniel Veillard      | Red Hat Desktop team http://redhat.com/
veillard redhat com  | libxml GNOME XML XSLT toolkit  http://xmlsoft.org/
http://veillard.com/ | Rpmfind RPM search engine http://rpmfind.net/