[dsdl-discuss] Re: Overlaying schema expressions

MURATA Makoto wrote:

>On Mon, 31 May 2004 12:36:52 +1000
>Rick Jelliffe <ricko@allette.com.au> wrote:
>
>
>
>>G. Ken Holman wrote:
>>
>>
>>
>>> But, in my understanding, a lot of people think that this is insane.
>>>
>>>Then it is time for us to proselytize, and I have a real-world
>>>situation where I will try to bring this up.
>>>
>>>
>>OTOH, the idea that there may be more than one schema per underlies XML
>>Schemas type derivation
>>by restriction, wildcards and substitution groups.
>>
>>
>
>Disagree. They are mechanisms for creating schema components by reusing
>diffrent schema components. In particular, when derivation by extension
>is used, an element valid against a derived type is not valid against
>the original type.
>
They are still different schemas. But derivation by extension and
especially <redefine>
show that W3C XML Schema allows multiple schemas for the same namespace.
It is the notion of multiple schemas in parallel that isn't part of the
deal.

I am in part responsible for <redefine> actually: part of Tim B-L's
requirements
for XML Schemas was that it would be useful for XHTML. Of course, there
was actually very little useful: indeed, XML Schemas was less useful than
DTDs at that stage because of type-mania. So we had to introduce <redefine>
very late to cope with changes to content models that did not fit into
derivation by restriction or extension: it clearly flags that there has
been a
breakdown in the (power of) the XML Schemas type system.

>
>
>>And I don't think
>>it is really so contentious that it is necessary sometimes to have
>>different orthogonal schemas, e.g.
>>anything + Schematron, to make up for shortcomings.
>>
>>
>
>I see nothing wrong in that scenario. But have people use W3C XML Schema
>together with Schematron?
>
>
>
Yes. I think it is hacky myself, but we get a lot of downloads of a free
extractor from our
site. Because no-one can use the PSVI yet, the fact that the Schematron
is not tightly coupled
to the WXS components is no problem. That will probably change.

Rick

>>Well, everything on a computer has to be expressed in syntax (or
>>something expressible in syntax). If you have this
>>absolute distinction between syntax and semantics, then "semantics"
>>reduces to "everything that cannot be expressed
>>on a computer" it seems to me. But things only have syntax because
>>they have semantics. So I don't really
>>understand the whole argument.
>>
>>
>
>Eri has already provided a very good explanation. He wrote: "Schemas provide a new
>indirect way to attach the applications that implement the semantics
>(the XSLT transformation in my example) to the labeled branch of the
> hierarchical tree. " I think that this indirection should not be mandated
>but rather left to additional layers such as data binding tools.
>
>CHeers,
>
>
>
>

--
DSDL members discussion list
To unsubscribe, please send a message with the
command  "unsubscribe" to dsdl-discuss-request@dsdl.org
(mailto:dsdl-discuss-request@dsdl.org?Subject=unsubscribe)