Daniel  Kulp wrote:
> 
>> 
>> Metro wsimport places all classes in the "rosette" folder, so we either
>> have a CXF bug in using "rosette_analysis" or a JAXB and Metro wsimport
>> bug in using "rosette".
> 
> I would say Metro/JAXB bug in using rosette and dropping the ".analysis"
> part.   
> I'm looking at the algorithm of Appendix D in the JAXB spec and really
> don't 
> see anywhere where it would drop that.   
> 

I think all three have a bug, actually.  A dot is a valid character in the
"NSS" as defined here: [1]

>From JAXB 2.2 (8 Jan 2009), section D.5.1 (and also JAXB 2.0 of 2006): 
   
2. Remove the trailing file type, one of .?? or .??? or .html.
The spec is defining a "file type" to mean .?? or .??? or .html, anything
else is not a file type.  (It is similar to Step #1, which defines the URI
schemes to be removed to be only "http" or "urn", even though more could be
considered uri schemes.)

//www.acme.com/go/espeak.xsd ==> //www.acme.com/go/espeak

Metro and JAXB are always treating everything after the dot as a file
extension and wrongfully removing it:
.xml -> removed (good)
.pumpkin -> removed (bad)

CXF is *never* removing the trailing file type :
.pumpkin -> not removed (good)
.xsd -> not removed (bad)
.html -> not removed (bad)

I think CXF has another bug:
given uri of "https:basistech.com:rosette.analysis"

Metro/JAXB treats https as part of the package (good):
https.basistech_com.rosette

CXF returns this error:
[INFO] org.apache.cxf.tools.common.ToolException: : is not a valid char in
the targetNamespace

So I think we have a JIRA ticket also for CXF.  (What class handles this?)

Glen

[1] http://xml.resource.org/public/rfc/html/rfc2141.html#anchor4
-- 
View this message in context: 
http://cxf.547215.n5.nabble.com/Missing-piece-of-urn-in-class-name-tp2112085p2255355.html
Sent from the cxf-user mailing list archive at Nabble.com.

Reply via email to