David Crossley wrote:

>Ross Gardler wrote:
>  
>
>>David Crossley (JIRA) wrote:
>>    
>>
>>>   [ 
>>>   http://issues.apache.org/jira/browse/FOR-675?page=comments#action_12332323 ] 
>>>
>>>David Crossley commented on FOR-675:
>>>------------------------------------
>>>
>>>There is a simple way to avoid these issues.
>>>
>>>Remove the xmlns attribute from every site.xml and make the following 
>>>change. The linkrewriter can then be used with the updated JXPath library.
>>>      
>>>
>>Is this is a robust workaround? I'm not intimate enough with the 
>>navigation generation to be able to predict the potential side effects 
>>of this.
>>    
>>
>
>I don't know either, but everything seems to be working.
>
>Haven't tried the views dispatcher yet.
>
>It is a drastic change, but this issue is too.
>
>One thing that it would affect is the future ability to have
>a new version of the namespace. I reckon that we should
>sacrifice that to get past this bug.
>
>  
>
>>(by the way, thank you David for keeping at this issue, I've been hiding 
>>from it ;-)
>>    
>>
>
>Thanks, i have been spending a huge amount of time on it.
>I have been deep in Cocoon input modules and JXPath,
>but i am not getting far. It needs someone with better
>Java/XML/Namespace/Cocoon foo.
>  
>
Yes, thanks David. But why you think we need deep cocon understanding to 
fix this bug? I am just asking. :-)

I believe, the problem is inside the commons-jxpath itself, as I posted 
in my latest report in jira. In short, 
http://issues.apache.org/bugzilla/show_bug.cgi?id=32360

To me the next step is forget cocoon for a while, create a simple 
testcase for the jxpath library check the error is there and fix it. 
Once the jxpath is fixed, import the patched lib to forrest. I bet it is 
going to work. ;-)

Unfortunately, for the lack of time recently, I was unable to follow do 
that. :-(

Best Regards,

Antonio Gallardo.