Re: Rationale for restricted uses of "all"

Thank you for your comments.  Personally, I share some of your concerns 
regarding the exact tradeoffs made in the extension/refinement space 
(though I do like the separation of the basic concepts).   One of the 
tricky things in doing a design like this is to know what to do when 
various constituencies strongly indicate they need a feature, as was the 
case with this single-inheritance, sequence-based form of extension.  As I 
sometimes have said, large committees tend to produce large languages.

Regarding systems like Schematron that can do more elaborate constraint 
checks and in that sense are more expressive:  they are very interesting 
systems, but I think Schematron in particular represents a significantly 
different point in the design/performance space.  I think I know how to 
optimize XML schema validation a lot more aggressively than I would 
suspect is possible with Schematron.  I could be wrong, but that's my 
intuition.  For high performance eCommerce scenarios, that's important.

I also agree in general on composability (is that a word?).  On the other 
hand, one thing that's made XML 1.0 an interop success is that there isn't 
much optionality in it.  You either understand well-formed or you don't. 
There are, of course, a few major choices such as whether you are 
DTD-aware and whether you validate.  But mostly, all processors handle all 
XML.  In the case of schema, the more we rely on composition to build the 
language, the greater the chance that I send you a schema that's perfectly 
legal, but that you can't handle (because I composed features that you 
don't support). 

I'm not sure we shouldn't wrap up this discussion, before it bothers the 
list with a huge discussion of all the tradeoffs made in the schema 
language.  Anyway, the above would be my reactions to the specific points 
that you've raised (and by the way, your little O'Reilly book on schema is 
very cool!)  Thanks.

------------------------------------------------------------------------
Noah Mendelsohn                                    Voice: 1-617-693-4036
Lotus Development Corp.                            Fax: 1-617-693-8676
One Rogers Street
Cambridge, MA 02142
------------------------------------------------------------------------







Eric van der Vlist <vdv@dyomedea.com>
Sent by: xmlschema-dev-request@w3.org
11/09/01 03:00 PM

 
        To:     xmlschema-dev@w3.org
        cc:     (bcc: Noah Mendelsohn/CAM/Lotus)
        Subject:        Re: Rationale for restricted uses of "all"


Hi Noah,

Noah_Mendelsohn@lotus.com wrote:

Thanks for sharing this insight with us!
...

> Just to avoid complicating an already overly large 
> specification, we decided to tackle extension initially only for 
> sequences, which models certain forms of OO single inheritance 
moderately 
> well. 


I think that this is the main point and what I find most questionable 

about W3C XML Schema is not the fact that it's not complex enough ;=) 

but rather the "usage" made of the complexity accepted into the

language...

Of course it's easy to criticise and most of what I have learned about 
schema languages have been learned studying the work of the Schema WG... 
but if I may be allowed one more critic, I would say that there are 2 
directions in which complexity can be deploied when defining a schema 
language: the first one is the expressive power (ie the scope of the set 
of instance documents which can be described) while the second are the 
"composing features" which let users recompose basic building blocs.

IMHO, the level of overall complexity would have been better used by 
focussing more on the expressive power than on the "composing features".

Expressive power is the "engine" or the basis of a schema language and 
you can't improve it afterward, while you can always add more "composing 
features" which I see more like a cherry on the cake...

Anyway, it's now the way it is ;=)  and for the best or the worse I 
guess we have to live with it for a while and try to use it for the best!

Thanks

Eric

-- 
Rendez-vous à Paris pour le Forum XML.
                    http://www.technoforum.fr/Pages/forumXML01/index.html
------------------------------------------------------------------------
Eric van der Vlist       http://xmlfr.org            http://dyomedea.com
http://xsltunit.org      http://4xt.org           http://examplotron.org
------------------------------------------------------------------------

Received on Friday, 9 November 2001 15:31:47 UTC