You'll probably need to put a breakpoint in the XMLStreamReader's
getCharacters or next() methods and get the stack trace at where the
whitespace blocks are returned. I don't think this is in CXF at all. I
think it's down in JAXB which there isn't much we can do about other than
create a small JAXB only testcase and log it with them.
Dan
On Wed February 3 2010 9:16:34 am Adam Funk wrote:
> On 2010-01-25, Daniel Kulp wrote:
> > Well, if anything, it's an issue either with the StAX parser or with
> > JAXB, not really CXF. Would you be in a position to create a small test
> > case using just JAXB?
> >
> > Basically, do something like: (pseudo code, probably not correct)
> >
> > JAXBContext ctx = JAXBContext.newInstance(TextWithNodes.class);
> >
> > XMLStreamReader reader = ... create a reader for you message
> >
> > Object o = ctx.createUnmarshaller().unmarshal(reader);
> >
> > And check out that object to see if that has the problem. In theory, it
> > should as that's pretty much all we do. The issue is whether the
> > reader is reporting the whitespace or not. You could also try
> > debugging that a bit:
> >
> >
> > while (int event = reader.next() != XMLStreamReader.END_DOCUMENT) {
> > System.out.println(event);
> > if (event == XMLStreamReader.TEXT) {
> > System.out.println("text: " + reader.getCharacters();
> > }
> > }
> >
> > or similar just to see if the reader is reporting the whitespace texts.
>
> Thanks. We did something like that. We know the XMLStreamReader is
> reporting the whitespace (since it's the same implementation that we
> use elsewhere, without losing it), so we think the DOMHandler is
> dropping the whitespace-only text nodes.
>
> Is there any way to configure it not to do this? Or can anyone
> recommend a more suitable DOMHandler?
>
--
Daniel Kulp
[email protected]
http://www.dankulp.com/blog
