RE: <model>- declaration

Good day, wherever you are :-)

first of all a brief meta-disclaimer: the disclaimer I used in my last
message was occasionally misunderstood. What I tried to say is: I do my
best to represent the current thinking of the group, but we can't afford
consciously polling for consensus for every message here. Consequently,
messages to this forum make no assertion about presence nor absence of
consensus, but my personal best effort to explain the current state of
the technology. XForms is work in progress anyway, so whatever I say
can't provide more than a snapshot (see also the standard disclaimer in
the Status of this document section of the XForms spec [1]). 

I hope the discussion here will remain on the highly technical level
that we all are used to instead of drifting into political guesswork.

Subject to this disclaimer, back to business: Roman, you are right. The
approach laid out mixes constraints with data. Still, matching the
constraint to the data by other mechanisms has turned out too expensive
for small devices.
Please remember that the second scenario in the previous message
supports full Schema validation and separation between meta-data and
data (even though it may not be available on all devices).

The example you provide mostly reflects the current state, except that
you have to explicitly prefix the required and readonly attributes so
that they fall into the XForms namespace.

For example:

<xf:instance>
	<notiz>
		<username xsi:type="xsd:string"
xf:required="true"/>
		<persoenlicheNotiz xsi:type="xsd:string" />
		<datum xsi:type="xsd:date" xf:readonly="true"/>
		<titel xsi:type="xsd:string" />
		<text xsi:type="xsd:string" />
	</notiz>
</xf:instance>

In order to integrate additional constraints beyond the Schema built-in
types, you will need a combination of an external Schema to define the
types, plus the xsi:type mechanism to attach the type to the element.
This is literally today's state of the discussion and could change. We
are working on use cases and examples as I write, and of course we'll
share them as soon as they are available. (There was an allusion to that
in my last message: "Further messages...")

One of the consequences of this architecture is that it will not be
possible in XForms Basic to attach types to attributes.

As always: Everybody, please let us know what you think so that we can
take your insight into account. Is an XForms Basic useful if it is
subject to the above restrictions?

Josef

[1] http://www.w3.org/TR/xforms

P.S.: Who here needs explanations of XML Schema terminology in order to
follow the discussion, and who is sufficiently fluent to proceed
without? 

> -----Original Message-----
> From: Roman Huditsch [mailto:r.huditsch@hico.com]
> Sent: Wednesday, July 25, 2001 9:14 AM
> To: Josef Dietl; www-forms@w3.org
> Subject: AW: <model>- declaration
> 
> 
> Good morning again!
> 
> 
> I would really appreciate the Case 1 of Josef's examples, but 
> this would
> IMHO mean that you mix up semantics with the expected structure of the
> result XML document. 
> In that case you would have to specify the datatypes in the instance
> model like this:
> 
> 	<xf:instance>
> 		<notiz>
> 			<username xsi:type="xsd:string"
> required="true"/>
> 			<persoenlicheNotiz xsi:type="xsd:string" />
> 			<datum xsi:type="xsd:date" readonly="true"/>
> 			<titel xsi:type="xsd:string" />
> 			<text xsi:type="xsd:string" />
> 		</notiz>
> 	</xf:instance>
> 
> Besides, will there be attributes like "maxlength" and 
> "minlength" like
> in HTML- Forms or how would such restrictions be implemented 
> in XForms? 
> 
> with best regards,
> Roman 
> 
> 
> 
> 
>  Roman Huditsch (RH ) 
>  
>  _____________________________________________________________________
>  hico Informations- und Kommunikations-Management Gesellschaft m.b.H.
>  TechLab, Thomas A. Edison Straße 2.
>  A-7000 Eisenstadt / Austria
>  phone: +43/2682/704-61-00; fax: +43/2682/704-71-61-10
>  e-mail:support@hico.com;  r.huditsch@hico.com
>  
> 
>

Received on Wednesday, 25 July 2001 10:57:35 UTC