Hi Werner,

Thanks for your time here.   I tried that property & properties file with 
the new attribute but had no change in results.  We'd certainly prefer to 
use the choice element as you'd assume, which would be mutually exclusive. 
 it actually looks as if setting maxOccurs at the choice level will write 
more than one type if they are provided in the xml

i.e., 

providing
...
        <foo>1</foo>
        <foo>2</foo>
        <bar>3</bar>
..
to 
.
         <xs:choice  maxOccurs="unbounded">
            <xs:element name="foo" type="xs:string" />
            <xs:element name="bar" type="xs:string" />
       </xs:choice>

yields:
 
2008-10-30 11:55:17,726 main DEBUG xml.UnmarshalHandler - #startElement: 
foo
2008-10-30 11:55:17,726 main DEBUG xml.UnmarshalHandler - #characters: 1
2008-10-30 11:55:17,726 main DEBUG xml.UnmarshalHandler - #endElement: foo
 
2008-10-30 11:55:17,726 main DEBUG xml.UnmarshalHandler - #startElement: 
foo
2008-10-30 11:55:17,726 main DEBUG xml.UnmarshalHandler - #characters: 2
2008-10-30 11:55:17,726 main DEBUG xml.UnmarshalHandler - #endElement: foo
 
2008-10-30 11:55:17,726 main DEBUG xml.UnmarshalHandler - #startElement: 
bar
2008-10-30 11:55:17,726 main DEBUG xml.UnmarshalHandler - #characters: 3
2008-10-30 11:55:17,726 main DEBUG xml.UnmarshalHandler - #endElement: bar


This above choice level behaiour is kind of a side note.  If you do have 
an opportunity to look into how we may succesfully set the element level 
of a choice without the maxOccurs of the choice, that would be a winner 
for us.


Best,

Jeff






Werner Guttmann <[EMAIL PROTECTED]> 
10/30/2008 03:38 AM
Please respond to
[email protected]


To
[email protected]
cc

Subject
Re: [castor-user] Re: problem with xs:choice order indicator






Hi Jeffrey,

I do remember that - now and then - issues have been reported related to
 unbounded choices and broken validation code. I will be having a
detailed look later on this week, but I simply do not have the time to
do this right now.

There's only one work-around I can suggest: can you - i a custom
castor.properties file - please override the following property:

# Property that allows to specify whether the validation for
# &lt;xs:integer&gt; should accept the old 'int/Integer' members
# as well; default to false.
#
org.exolab.castor.xml.lenient.integer.validation=true

and see whether that makes a difference ?

Regards
Werner Guttmann

[EMAIL PROTECTED] wrote:
> ----- Forwarded by Jeffrey Kramer/USA/DDS on 10/29/2008 04:53 PM -----
> 
> 
> 
> 
> Hi Werner,
> 
> I should have mentioned initially this is Castor version 1.2 on a 
windows 
> xp machine with java version 1.5.0_16. 
> 
> It looks like the major question here is how Castor is handling the 
> attributes of the Choice elements & it's respective nested simple 
elements 
> 
> 
> under the following assumption:
> 
> The expression <xs:choice maxOccurs="unbounded"> allows the contents to 
be 
> repeated one or more times. 
> The expression <xs:element name="X" maxOccurs="unbounded"> allows the 
> single element to be repeated one or more times
> 
> 
> For xml containing: 
> 
>         ...
>         <foo>1</foo>
>         <foo>2</foo>
>         <office>NY</office>
>         ...
> 
> I've observed the following cases:
> 
> 
> (1) maxOccurs attribute set in choice element
> 
>   <xs:choice maxOccurs="unbounded">
>             <xs:element name="foo" type="xs:string" minOccurs="1" />
>   </xs:choice>
> 
> 
> => works without error
> 
> (2) Here the case where the choice element does not have a maxOccurs 
> attribute set ( default = 1)
> 
>   <xs:choice >
>             <xs:element name="foo" type="xs:string" minOccurs="1" 
> maxOccurs="unbounded" />
>   </xs:choice>
>   <office>NY</office>
> 
> => throws the following exception:
> 
> -- org.exolab.castor.xml.MarshalException: Element with name foo passed 
to 
> type companyEmployee in incorrect order; expected element with name 
> 'office' or any other optional element declared prior to it.{File: [not 
> available]; line: 9; column: 7} 
> 
> This 2nd case is actually a format we're interested in using.   Any 
> insight or suggestion you may be able to provide would be greatly 
> appreciated.
> 
> Thanks,
> 
> Jeff
> 
> 

---------------------------------------------------------------------
To unsubscribe from this list, please visit:

    http://xircles.codehaus.org/manage_email



Reply via email to