ode-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Karolis Petrauskas <k.petraus...@gmail.com>
Subject Re: XML namespaces
Date Tue, 01 Feb 2011 23:48:09 GMT
Hi Tammo,

    Thank you for your reply. I will open JIRA ticket with my patch
attached. It looks like my patch also relates to

Karolis Petrauskas

On Tue, Feb 1, 2011 at 3:50 PM, Tammo van Lessen <tvanlessen@gmail.com> wrote:
> Hi Karolis,
> sorry for the belated reply. What you describe is a indeed a bug. From
> your text I don't understand what your approach exactly is, what about
> opening a JIRA and attaching your patch so that we can review it?
> Patches and contributions always welcome :)
> Thanks,
>  Tammo
> On Wed, Jan 5, 2011 at 23:09, Karolis Petrauskas <k.petrauskas@gmail.com> wrote:
>> Hi all,
>>    I'm using Apache ODE in my project to run processes. Inheritance
>> of complex types is heavily used in the XML schemes of the web
>> services. As a consequence, xsi:type attributes are used in the
>> messages extensively. Here I faced some problems. It seems ODE tries
>> to remove XML declarations of namespaces, that are not used in the
>> document. This optimization unfortunately does not take xsi:type
>> attributes into consideration. As a consequence, message like this
>> (pseudo code):
>>    <message xmlns="messageNS">
>>        <order xsi:type="o:MyOrder" xmlns:o="orderNS"
>> xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"/>
>>    </message>
>> converts to something like this (pseudocode again):
>>    <message xmlns="messageNS">
>>        <order xsi:type="o:MyOrder"
>> xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"/>
>>    </message>
>> i.e. xmlns:o="orderNS" is removed, as the prefix "o" is not referenced
>> anywhere except the contents of the xsi:type attribute. Resulting
>> document is not valid anymore if complex type "MyOrder" was derived
>> from an abstract complex type.
>>    I have fixed this behavior and it works at least for my process.
>> The side effect is that a lot of unneeded namespace declarations are
>> now passed around. In my case it is better than corrupted messages. I
>> want to contribute my patch to ODE, but I'm not shore, if my approach
>> is OK and would like to get your opinion about this issue.
>> Looking forward for your comments,
>> Karolis Petrauskas
> --
> Tammo van Lessen - http://www.taval.de

View raw message