RE: [xml-dev] xsd:anyType and xsd:any

[noah_mendelsohn@us.ibm.com]

Shlomo Yona writes:

> OK. So, both xsd:any and xsd:anyType refer to
a subtree. While the 
> default parseContents for xsd:any is "strict" the parseContents
for 
> xsd:anyType is "skip"?

I'm afraid you are still confused.  As
I said before, one of these is a "type".  The architectural
purpose of a type in XML schema is to be a constraint on what you're calling
a subtree, I.e. independent of the name of an element, what are its legal
contents.  

xsd:any is not a type, but it's usable
as a term of a particle.  It doesn't directly constrain a subtree
at all.  Particles are things like the references to elements A and
C in the following.

        <xsd:complexType name="T">
        <xsd:sequence>
                <xsd:element
ref="A"/>
                <xsd:any />
                <xsd:element
ref="C"/>
        </xsd:sequence>
        </xsd:complexType>

You will note that these references are to the >name<
of an element (will, literally to the declaration).  The particles
in the sequence, in this example, tell what element >names< may appear
as >immediate children< of the element of type T.   They also
do get you started on validating those elements, because they get you to
the element declarations, which in turn have types, which in turn constrain
the contents of those elements, perhaps by having groups such as sequences,
which have particles....and the whole thing is thus recursive.  The
<xsd:any> here is just a special type of particle that matches any
name.  >It is NOT a type!< There are controls you can put on
the xsd:any to make it match only names in certain namespaces, etc., as
I assume you've read in the schema Recommendation [1].

There is no such thing as parseContents, but there
is processContents on the wildcards.  Since the wildcard can match
elements of any name, there's a possibility that some such elements will
have (global) declarations in the schema and some won't.  In case
that's of concern to the schema author, processContents merely tells whether
it's an error not to have such declaractions, whether they are to be ignored
even if present, or whether they are to be used if available to validate
the content, but not cause an error if missing.

As I said in my first note, the built in type named
xsd:anyType behaves as if it had a wildcard in it for elements (and another
for attributes, by the way.)  anyType is magic in some ways, but it's
a useful approximation for some purposes to assume the wildcard is there.
 As you will see at [2], the Recommendation says:

"There is a complex type definition nearly equivalent
to the ·ur-type definition· present in every schema by definition. It has
the following properties:"  (in this, ur-type definition means
xsd:anyType -- don't ask!)  If you look at that, you will see that
its particles property is defined as:

{particles}
a list containing one particle with the following properties:

Property

Value

{min occurs}

0

{max occurs}

unbounded

{term}

a wildcard with the following properties:

Property	Value
{namespace constraint}	any
{process contents}	lax

So, there's your answer:  it's lax.  (There
is something funny going on there, because this is the root of the type
hierarchy, yet lax will reject some things that skip will accept.  We
had extensive discussions for Schema 1.1 about doing something to fix that,
but my somewhat vague recollection is that we didn't come up with anything
sufficiently satisfactory to propose as a change.  So, that's a glitch
if I'm remembering correctly.)

Noah

[1] http://www.w3.org/TR/2004/PER-xmlschema-1-20040318/#Wildcards
[2] http://www.w3.org/TR/2004/PER-xmlschema-1-20040318/#d0e9210

--------------------------------------
Noah Mendelsohn 
IBM Corporation
One Rogers Street
Cambridge, MA 02142
1-617-693-4036
--------------------------------------

"Shlomo Yona" <S.Yona@F5.com> 03/20/2007 12:29 PM

To:        <noah_mendelsohn@us.ibm.com>         cc:        <xml-dev@lists.xml.org>         Subject:        RE: [xml-dev] xsd:anyType and xsd:any

Hello, and thanks for your detailed answer.
Actually, reading the references that you listed got me into this confusion...

OK. So, both xsd:any and xsd:anyType refer to a subtree. While the default
parseContents for xsd:any is "strict" the parseContents for xsd:anyType
is "skip"?

How can I tell the default parseContents of such "joker" or "wildcard"
definitions?

Thanks.

Shlomo.




-----Original Message-----
From: noah_mendelsohn@us.ibm.com [mailto:noah_mendelsohn@us.ibm.com] 
Sent: â 20 îøõ 2007 17:36
To: Shlomo Yona
Cc: xml-dev@lists.xml.org
Subject: Re: [xml-dev] xsd:anyType and xsd:any

Shlomo Yona asks:

> I was not able to fully understand the difference between an xsd:any
> and an element with a type xsd:anyType.

Very different.  xsd:anyType is a type, like xsd:integer (though 
xsd:anyType is special in that it can act as a simple or complex type,
and 
it places essentially no restrictions on the tree that it validates --
thing of it loosely as the Schema language's analog of java.lang.Object).
A sample use would be:

        <xsd:element name="e" type="xsd:anyType"/>

This would mean that elements named <e> can have any content, any
attributes, etc.

xs:any is a wildcard, usable as a term in a content model.  For example:

        <xsd:complexType name="T">
        <xsd:sequence>
                <xsd:element
ref="A"/>
            <xsd:any />
                <xsd:element
ref="C"/>
        </xsd:sequence>
        </xsd:complexType>
 
Elements of type T must have content <A/><???/><C/>,
where <???> can be 
any named element.  Now, if you look really closely there is an 
approximation to the definition of xsd:anytType given for reference in
the 
Recommendation, and it uses an xsd:any wildcard as the means of saying
that it allows any elements. 

I hope you'll see that most of your questions are answered once you get
the above straight.  I would encourage you to carefully read the schema
primer [1], or some of the good books on W3C XML Schema (Priscilla 
Walmsley's is among the one's that's very well respected. [2]).

Noah

[1] http://www.w3.org/TR/xmlschema-0/
[2] 
http://www.amazon.com/Definitive-XML-Schema-Priscilla-Walmsley/dp/0130655678/ref=sr_1_1/103-2086530-8739813?ie=UTF8&s=books&qid=1174404823&sr=8-1

--------------------------------------
Noah Mendelsohn 
IBM Corporation
One Rogers Street
Cambridge, MA 02142
1-617-693-4036
--------------------------------------








"Shlomo Yona" <S.Yona@F5.com>
03/20/2007 11:08 AM
 
        To:     <xml-dev@lists.xml.org>
        cc:     (bcc: Noah Mendelsohn/Cambridge/IBM)
        Subject:        [xml-dev]
xsd:anyType and xsd:any


Hello,
 
I was not able to fully understand the difference between an xsd:any and
an element with a type xsd:anyType.
 
Is there any difference?
Do both wildcard methods refer to any child element(s) or does one or both
refer to a wildcard subtree?
What should be the parse mode (parserContents) for processing xsd:anyType?
Should is be same as xsd:any’s “skip” or “lax” or “strict” or perhaps
to 
something else?
The section 3.4.7 Built-in Complex Type Definition (in 
http://www.w3.org/TR/xmlschema-1/) says, if I understand correctly, that
the subtree should be processed in “lax”. Is that so?
What’s the difference, if any between the following wildcard definitions
(with regards to the depth of subtree they refer to, the processing mode
and any other issue):
·         xsd:any
·         element definition with type=”xsd:anyType”
·         a complexContent that has a restriction base=”xsd:anyType”
with 
an xsd:anyAttribute namespace=”##other” and processorContents=”lax”
·         a complexContent that has a restriction base=”xsd:anyType”
with 
an xsd:anyAttribute namespace=”##other” and processorContents=”skip”
Thanks.
 
Shlomo.