W3C home > Mailing lists > Public > xmlschema-dev@w3.org > December 2001

RE: Expressing varying cardinalities of unordered elements in a complex-type

From: Mason Lee <mgl@netspace.org>
Date: Mon, 3 Dec 2001 20:02:23 -0500 (EST)
To: "Eddie Robertsson" <eddie@allette.com.au>, <xmlschema-dev@w3.org>
Message-ID: <PGEEIHHNNCEOPCHBJONJEEFCCMAA.mgl@netspace.org>
  Oh.  Bummer.  Well, thanks, Eddie.  I'll look into the other schema
checkers you mention, but in my case it'll probably be easiest to use the
unrestricted choice group you mention and validate the uniqueness of the
couple of singleton elements manually in my app after the schema check.
  I was hoping I could use a simple XMLSchema alone as the formal definition
of my document's required structure, but I think that having to reference
external checkers would make the schema too ugly for that, and I'd be better
served by just writing some documentation describing the restrictions.
  Out of curiousity, does anyone have any links to discussions as to why
this sort of constraint wasn't included in the W3C XML Schema?  Or more
specifically, why "all" doesn't allow for cardinality >1?

Cheers,
Mason



-----Original Message-----
From: Eddie Robertsson [mailto:eddie@allette.com.au]
Sent: Monday, December 03, 2001 4:07 PM
To: Mason Lee
Cc: Jeni Tennison; xmlschema-dev@w3.org
Subject: Re: Expressing varying cardinalities of unordered elements in a
complex-type


Hi Mason,

>   Not being used to functional programming and finding XML Schema somewhat
> elusive, I just have to ask again then, a slightly different question:
>   How might I use XML Schema to define a complex-type of 1 Element_A, 1
> Element_B, 0 or 1 Element_C, and any number of Element_D, all appearing in
> any order?
>   Thanks in advance.  I'm not sure why this seemingly simple situation
> doesn't appear in any of my books.

The answer to why it doesn't appear in any of your books is because you
can't
express these constraints with W3C XML Schema. As Jeni has shown you can
play
around with mixing the different groups to achieve some of the more simple
interleave constraints and in many cases an unrestricted choice group is
enough.

<xs:choice minOccurs="0" maxOccures="unbounded">
   ...element declarations...
</xs:choice>

To declare the content model you describe above you have to move to a
different
schema language. Both RELAX-NG and Schematron can express the content model
you
describe and my personal favourite is to use Schematron rules embedded in
W3C
XML Schema. See [1] for an example of how to use embedded Schematron rules.

Cheers,
/Eddie

[1] http://lists.w3.org/Archives/Public/xmlschema-dev/2001Nov/0049.html



>
>   Regards,
>   Mason
>
> -----Original Message-----
> From: Jeni Tennison [mailto:jeni@jenitennison.com]
> Sent: Saturday, December 01, 2001 5:39 AM
> Subject: Re: Order of elements in a complex-type
>
> Hi Mason,
>
> > Hi all! I saw Erik Beijnoff's good answer to the above question in a
> > recent thread. I wonder further, however, how could this complex
> > type be defined such that the elements "TextBox" and "TextArea" may
> > appear any number of times, but the element "Header" must appear
> > only once, and while retaining the interesting characteristic that
> > any of these three clild elements may appear in any order. Any
> > ideas?
>
> Assuming that you don't care if the TextBox and TextArea elements are
> intermingled, and there's no minimum number of occurrences of TextBox
> and TextArea, you could use the following:
>
> <xs:complexType name="FormType">
>   <xs:sequence>
>     <xs:group ref="TextBoxOrTextArea" minOccurs="0"
>               maxOccurs="unbounded" />
>     <xs:element name="Header" type="HeaderType" />
>     <xs:group ref="TextBoxOrTextArea" minOccurs="0"
>               maxOccurs="unbounded" />
>   </xs:sequence>
> </xs:complexType>
>
> <xs:group name="TextBoxOrTextArea">
>   <xs:choice>
>     <xs:element name="TextBox" type="TextBoxType" />
>     <xs:element name="TextArea" type="TextAreaType" />
>   </xs:choice>
> </xs:group>
>
> (Using a group so that you only have to declare TextBox and TextArea
> once.)
>
> Cheers,
>
> Jeni
>
> ---
> Jeni Tennison
> http://www.jenitennison.com/
Received on Monday, 3 December 2001 20:38:29 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Tuesday, 11 January 2011 00:14:25 GMT